MEASUREMENT OF THE EFFECT OF INTERNET BANDWIDTH ON SCIENTIFIC JOURNAL ACCESS

• Main Author: Danial
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/36858
• Institution: Institut Teknologi Bandung

The internet bandwidth needs at ITB are increasing. This is also related to the amount of costs incurred by ITB to provide the internet bandwidth. The absence of management of internet usage data at ITB that relates to user behavior. At present the internet user behavior at ITB is not managed yet. So it is not known exactly how much bandwidth needs are mainly for the process of carrying out research and teaching at ITB. Purchasing internet bandwidth is based on the number of usage of previous years, not based on the modeling of the number of needs and the use of existing bandwidth capacity. The research objective is to measure ITB internet bandwidth requirements based on user groups. Then look for patterns of academics in using internet bandwidth at ITB. The research was conducted using two stages, namely data mining and clustering. Data mining consists of the stages of Data Cleaning, Data Integration, Data Selection, Data Transformation, Data Mining, Pattern Evaluation, Knowledge Presentation. Then clustering was carried out based on the K-Means, Agglomeration and KMedoids methods. The process of cleaning data on raw data must be done as well as possible, because it will determine the quality of the data to be processed later. The variables used are the username, the amount of data used, and the content visited. The next stage is to look for data that accesses scientific / journal related content. Then look for data that accesses entertainment content. Then the aggregation of the amount of data used is based on the username to get the amount of data usage in general, aggregation based on username and content to get the number of usage based on username and content. The content group is divided into three types. Scientific content is a group of content that contains scientific material, including journals. Entertainment content groups are groups that contain entertainment content such as social media and youtube and similar content. General groups are groups that contain content other than the two previous groups. After obtaining the above aggregations, a clustering based on the K-Means, Agglomeration and KMedoids methods was carried out. The results obtained iv illustrate that there are three clusters of users on the ITB internet network. Namely the scientific content cluster (cluster 1), entertainment access cluster (cluster 2), email content cluster and general content (cluster 3). Cluster 1 is a cluster that accesses entertainment content and / or general content but is large enough to access scientific content compared to other contents. Cluster 2 is a cluster that accesses very little scientific content compared to entertainment content and or general content. The third cluster is a cluster that only accesses entertainment content and or general content. In the results of the study, cluster formation in the K-Means method was better than the Agglomeration method and KMedoids. The cluster results in the KMeans method provide a consistent comparison between the amount of access to scientific content compared to other content. For the Agglomeration method, it can be easily seen that users do not access scientific content but are included in cluster 1. For the KMedoids method, cluster 2 is formed so small that it is not representative compared to other clusters.